The stepped wedge cluster randomised trial workshop: Session 2

Introduction: what is the SW-CRT?

Defining features, common variations, and some salient

examples

Karla Hemming

30/08/2016

Why do we need another method of

evaluation?

Evaluation of policy and service delivery

interventions

• The Matching Michigan Study– Secular trend

– Intervention effect

• The Oregon Experiment – Randomised

– No primary outcome

• Mexican universal health

insurance– 74 matched clusters

– Staggered implementation

Evidence based policy interventions

Working constraints

• Stakeholder’s desires

• Pragmatic limitations

• A priori beliefs

How the SW-CRT can help

• All clusters ultimately get

intervention

• Sequential roll out

• Robust evaluation

What is the SW-CRT?

The Stepped Wedge Cluster Randomised Trial

Example 1: A cross-sectional SW-CRT

The EPOCH study – a cross sectional

SW-CRT

• Evidence based integrated care

pathway (ICP)

• Emergency laparotomy

• Setting:

– 90 hospitals

• Outcome:

– 90 day mortality

• Sample size:

– 27,500 patients

– 90% power to detect a change

from 25% to 22%

• Routinely collected outcome

– No consent

Pearse R, Pedan C, Bion J, Faiz O, Holt P, Girling A, et al. HS&DR - 12/5005/10. Enhanced Peri-Operative Care for High-risk patients (EPOCH) trial: a stepped

wedge cluster randomised trial of a quality improvement intervention for patients undergoing emergency laparotomy [protocol]. 2013.

www.nets.nihr.ac.uk/projects/hsdr/12500510.

Example 2: A cohort SW-CRT

Depression management– a cohort

SW-CRT

• Structural multidisciplinary approach

to depression management

• Participants – Residents who provided informed consent

• Setting:– 33 units within nursing homes (Dementia /

Somatic)

• Outcome:– Depression prevalence

• Sample size:– 793 patients

– 80% power to detect a 30 to 40% reduction

in prevalence (circa 20-30%)

• Outcome– Questionnaire (Cornell scale for

depression)

Leontjevas R, Gerritsen DL, Smalbrugge M, Teerenstra S, Vernooij-Dassen MJ, Koopmans RT. A structural multidisciplinary approach to depression

management in nursing-home residents: a multicentre, stepped-wedge cluster-randomised trial. Lancet. 2013 Jun 9;381(9885):2255-64. doi:

10.1016/S0140-6736(13)60590-5. Epub 2013 May 2. PubMed PMID: 23643110.

Example 3 – the Gambia hepatitis

study

The Gambia hepatitis study

• Step lengths 10 to 12 weeks

• National coverage after 4 years

• Geographically defined areas

• Study started 1980s

– 30 year follow-up

• Vaccine

– efficacy against hep B

• Main outcome

– liver disease

• Vaccine rolled out

– national infant

vaccination schedule

Is the SW-CRT the right design for my

trial?

• Pragmatic considerations

– Does it allow a randomised evaluation which otherwise would

not be possible?

• Logistical considerations

– Allows sequential role out

• Efficiency

– Minimise number clusters / participants / observations

• Duration

– Will it necessarily extend the trial?

– Time between exposure and follow-up

Word of caution …

• Lack of concealment

of allocation

– Risk of selection bias

• Avoid individual

patient recruitment

– Routinely collected

outcome data

Chalmers:

“Although one of the reasons that the streptomycin trial has become iconic is …

random number tables …. it was because successful concealment of allocation”

Example of a CRT with lack of

concealment of allocation

• Results in baseline

imbalance

• Due to recruitment of

individuals after

allocation known

• Will it be better or

worse in SW-CRT?

Variations to the common design

• More than two treatment comparisons

• Hybrid designs – mixtures of parallel designs and stepped studies

• Transition periods

• Multiple levels (i.e. clustering within clustering)

• Repeated measures (cohort designs)

To follow….

• What sample size do I need?

• What is the most efficient design?

• How to extend the design to all for more than two treatments?

• How do I analyse a SW-CRT?

Summary

• SW-CRT a pragmatic study design which reconciles the need for

robust evaluations with political or logistical constraints.

• Unbiased design when:

– No individual patient recruitment (routinely collected outcome)

• Efficient design when:

– Higher the ICC (process outcomes)

– Limited number of clusters

• Design and analysis

– Appropriate consideration of time effects in power and analysis

References 1 (methodology)

• Mdege ND, Man MS, Brown CATN, Torgerson DJ. Systematic review of stepped wedge cluster

randomized trials shows that design is particularly used to evaluate interventions during routine

implementation. Journal of Clinical Epidemiology 2011; 64:936–948.

• Hemming K, Haines TP, Chilton PJ, Girling AJ, Lilford RJ. The stepped wedge cluster randomised

trial: rationale, design, analysis, and reporting. BMJ. 2015 Feb 6;350:h391. doi: 10.1136/bmj.h391.

PubMed PMID: 25662947.

• Hussey MA, Hughes JP. Design and analysis of stepped wedge cluster randomized trials.

Contemp Clin Trials. 2007;28(2):182-91.

References 2 (motivating examples)

• Bion J, Richardson A, Hibbert P, Beer J, Abrusci T, McCutcheon M, et al. ‘Matching Michigan’: a

2-year stepped interventional programme to minimise central venous catheter-blood stream

infections in intensive care units in England. BMJ Qual Saf2013;22:110-23

• Baicker K, Taubman SL, Allen HL, Bernstein M, Gruber JH, Newhouse JP, Schneider EC, Wright

BJ, Zaslavsky AM, Finkelstein AN; Oregon Health Study Group, Carlson M, Edlund T, Gallia C,

Smith J. The Oregon experiment--effects of Medicaid on clinical outcomes. N Engl J Med. 2013

May 2;368(18):1713-22. doi: 10.1056/NEJMsa1212321. PubMed PMID: 23635051; PubMed

Central PMCID: PMC3701298.

• King G, Gakidou E, Imai K, Lakin J, Moore RT, Nall C, Ravishankar N, Vargas M, Téllez-Rojo

MM, Avila JE, Avila MH, Llamas HH. Public policy for the poor? A randomised assessment of the

Mexican universal health insurance programme. Lancet. 2009 Apr 25;373(9673):1447-54. doi:

10.1016/S0140-6736(09)60239-7. Epub 2009 Apr 7. PubMed PMID: 19359034.